Automated Criminal Identification System Using Face Generation

Abstract

The goal of the project is to aid the police department to ease the process of criminal recognition. We first initiate the recognition by the generation of face images based on image generation by using their sketch and visual attributes. The images we are generating are hyper-realistic in nature. This is followed by facial identification based on the image generated in the previous phase. The Model we have chosen here is a Generative Adversarial Network. The model learns the association by the use of a generator and a discriminator. We are using a context based Generative Adversarial Network to do the same. Then, we are running the process of facial identification for recognizing the person from the image generated with the help of facial recognition. For this we use the HOG Face recognition algorithm. We work towards aiding the police department in reducing their sketching time and criminal recognition time.